[DRBD-user] drbd questions

Ben Beuchler insyte at gmail.com
Mon Mar 1 08:07:56 CET 2010

Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.

> 1. meta-disk internal means i do not have to take care about it because it
> will be created automatically based on drbd-size, al-extends and sync-rate?

If you are adding DRBD with internal meta-data to an existing device
with data in place, then you do need to increase the size of the

(Note especially the large "Caution" section.)

> 2. meta-disk internal will be automatically created at the end of the
> partition - so, on raid hardware you can increase the storage on the
> underlaying device by adding more disks and grow the partition, but what
> does this mean for drbd?
> when the partition will grow and the meta-disk stay on its position - will
> it be overwritten after some time?

No, because the filesystem (and the actual block device made available
by DRBD) will always be smaller than the underlying block device.
DRBD is mediating access to the underlying block device, so it will
not accidentally overwrite its own metadata.

> or will it automatically move to the end of the partition, or is it
> generally not possible to increase a drbd-device and should be rather build
> new?

Growing DRBD resources is easy and well-documented:


> 3x gbit server nic's bonding mode 0 with 2.15 gbits/sec throughput - jumbo
> frames on for syncing only (iperf)

Bonding NICs for throughput performance is generally not recommended.
As all of your data is transferred over a single TCP stream, packets
arrive on the remote end out of order, requiring significant overhead
to fix.  As I understand the testing Linbit has done, 2 bonded NICs
will provide *slightly* better performance than one, but three
actually performs worse than a solitary NIC.


More information about the drbd-user mailing list